A Novel Human Motion Capture System Based On Inertial Sensor

Human motion information has been applied in a wide range of areas,such as business,virtual reality games,and medical rehabilitation services.In the filed of medical rehabilitation,human motion information could help doctors to obtain an objective perspective of the movement conditions in patients with hemiplegia after stroke.Thus personalized training plans could be formulated so as to enormously enhance the quality and the results of rehabilitation.Therefore,human motion information has great application value.At present,several common methods are used to capture human motion.For example,methods that are based on mechanical,optical,acoustic,electromagnetic,and video-based systems.In general,due to its rigid connection and non-flexcible structure,mechanical systems pose movement constraints of the testing subjects.Optical systems are expensive and require non-reflective field for image acquisition.Acoustic systems rely on the time differences of acoustic waves and are prone to environmental inteference.Electromagnetic systems orientate by magnetic filed so the testing environment should not exist any metals.Video-based capturing is very difficult to achieve.In order to overcome the above-mentioned shortcomings of common motion capture systems and to achieve accurate human motion capture,this paper proposes a convenient,low-constraint,wearable capture system based on the theory of strapdown inertial navigation and biomechanics.The system acquires real-time human motion information through inertial sensor nodes worn on the corresponding human joints.The set of inertial sensors is composed of accelerometers,gyroscopes and magnetometers with 9 axes to ensure the complexity of the data acquired.Node data are obtained through a micro-control unit and the quaternion is updated by low-pass filtering and Kalman filtering.The pre-processed data are then packed in agreement with the data format of the host.The packaged data are sent to the host computer through wifi and are unpacked to control thereal-time motion of a virtual human model to achieve final human motion capture.According to human biomechanics,in order to achieve the least mutual influence of human joints in motion and to ensure real-time data acquisition rate,local wired connection and whole-body wireless connection are proposed in this paper for data transmission.Meanwhile,three joints connected together are designed as a whole based on the design module.Such design could rapidly expand from three joints to the whole body once the first model is verified.Therefore,this paper focuses on the design and verification of an arm motion capture system.Single joint is studied according to its biomechanical characteristics.Initialization algorithms are corrected for the initial position of the arm.Afterwards,three joints are combined and joint distortion under the same coordinate system is solved through normalization algorithms.And the overall design is completed.Experiment results show that wearable sensors could be utilized for real-time human motion capture.This paper provides new prospects for future design and has innovative and application value.